Inconsistent wall thickness

I am struggling with this one for quite a while now and I am now seeking for help here.
I have done e-step calibration, optimized the flow rate, performed a retraction, linear advance and PID calibration but have yet to find the reason for my inconsistent wall thickness. I have also tried PETG and PLA Pro at different extrusion temperatures without any change. And I exchanged the heatbreak and nozzle and made sure the hotend temperature sensor is deep inside the heat block. Still, all the same. Lowering the maximum acceleration or using slower print speed gave me some minor improvements but I kind of doubt that this is the root of my issue. Here you can see the results I am getting:

The Cura project and GCode file: wall-thickness-test-PlaPro.zip (434.3 KB)

How can I get more consistent results?

P.S.: My machine specific additions to the GCode are:

M201 X1000 Y1000 Z100 E10000 ; sets maximum accelerations, mm/sec^2
M425 F1 S0 X0.04 Y0.06 Z0 ; Backlash Compensation
M92 E235.50 ; Set new extruder steps/mm
M900 K0.055000 T0 ;added by LinearAdvanceSettingPlugin

Everything looks good to me. I donā€™t see any red flags in what youā€™ve posted that would explain anything useful.

Do you have wall thickness variation on a single-wall (vase mode) calibration cube?

Is it related to the curve in the wall? Or does it vary on straight sections too.

If itā€™s only happening on curves then itā€™s probably tied to acceleration in the form of linear advance, would be my guess.

Iā€™m also assuming the variation in thickness is constant for each Z layer and this is not a filament with varying diameter.

If you rotate the part on the print bed 90 degrees does the variation stay the same? If you rotate it 45 degrees does it stay the same? Is this tied to one or 2 axes having motion issues?

Have you attempted to see if the linear modules are moving uniformly and you donā€™t have a lead screw wobble or pitch variation?

Is there any discernible play in the carriages that could introduce mechanical shifts that the backlash compensation is not removing?


I am most interesting in what happens if you print a single walled vase aligned at also maybe at 45 deg to the print bed.

Iā€™d also be interested in a single walled oval shape to see if introducing curves affects things.

1 Like

It is related to changes in the movement direction.

Yes, consistent filament diameter and same result in each layer.

Same results with 45 or 90 degrees rotation (I made sure the seam is at the same point).

I used a digital dial gouge to measure the backlash of the linear modules for X and Y axis. You can find the correction values above. I cannot see any issue here. I have very clean lines in X and Y direction over a long path.

The results are the same regardless of the orientation and placement. So I guess not.

Do you have any sample file I can try out? What results do you expect?

My personal opinion is that the viscosity of the material is not modeled correctly. But I havenā€™t heard much about such issues and my K value is calibrated. So I am bound to believe that this is mostly my issue. Or do you get similar results?

P.S.: I also tried PrusaSlicer instead of Cura and got the same results.

As @brent113 suggested, based on your image it seems to be tied to areas where acceleration/deceleration is taking place and so may be linked to linear advance. Linear advance is essentially a flow compensation that is applied when the print head is accelerating/decelerating in an attempt to achieve more even extrusion across a range of speeds.

Below is a post I made on this but if you search ā€œlinear advanceā€ or ā€œk valueā€ on the forums im sure there will be a wealth of information on how to calibrate this.

Thank you for pointing this out. However, I have already done this many times and always get the same result: 0.055 for PLA (as seen in my original post). I used this tool to generate the needed GCode with 25mm/s and 60mm/s.

Any other idea? Do you get different results with that object?

Unfortunately my printer is currently packed away and so I cannot test this file personally. I may be wrong but the variation in thickness in your image to me looks like it could correlate with too low a k value. Personally, I would run a quick test print with higher K value and slightly lower accel just to rule it out:

M900 K0.07
M201 X750 Y750 Z100 E10000

1 Like

Did you check that the option ā€œautomatic line widthā€ or something similar is turned off?

I agree with the posted above.

Try to print a calibration cube in vase mode.

I just tried K0.07 and got more or less the same results.

I could not find any such option in Cura.

That will be my next step but the test I did already used a solid object and slicer with single wall mode (not vase mode)ā€¦

The calibration cube looks good I would say. The wall thickness is 0.48Ā±0.02mm.
Or is there anything else I can read from this except the elephant feet?


I redid my original object in vase mode without the holes. The result is the same as before.


You can clearly see the change in the wall thickness at the change of the print direction.

I made a little video of the process. I am not sure but the print speed does not looks uniform to me:

Here the files I used for printing:
round-tolerance-0.02-solid.stl (113.8 KB)
round-tolerance-0.02-solid.gcode (602.3 KB)

Print this model with luban to check if its a slicer issue.
After this i would print the model with various k-factors.
Maybe also tuning your junction deviation would help but try the other things first.

I already tried PrusaSlicer to rule out a slicer issue (as Luban uses Cura as backend anyway).
Setting junction deviation from M205 J0.01 to M205 J0.05 did not make any visual or measurable changes here. So I did some more linear advance calibration. First I tried K0.05 to K0.25 with my test object. The best value here seemed to be around 0.1. I redid the marlin linear advance calibration with values from 0.01 to 0.16 and 25mm/s to 60mm/s. But this time I used the pattern type alternate instead of standard and focused on the corners rather than the straight lines. I got 0.11 instead of 0.055 i.e. twice my original value. A reprint did turn out better than before. So I guess the speed change from 25mm/s to 60mm/s is not enough to simulate cornering events. I will settle with K0.11 for now. Letā€™s see how other objects turn out with this setting.

Thank you for the hints @xchrisd

1 Like

Inconsistent wall canopies thickness can be caused by various factors, and you seem to have already tried many of the common solutions. However, here are some other suggestions that may help:

  1. Check for mechanical issues: Make sure that your printerā€™s frame is sturdy and that there are no loose screws or belts. Check that the X, Y, and Z-axis movement is smooth and consistent, and that there is no binding or slipping during printing.
  2. Calibrate your extruder: Verify that your extruder is calibrated accurately and that it is providing the correct amount of filament. This can be done using a calibration cube or by measuring the extruded filament with a caliper.
  3. Adjust your extrusion multiplier: Try adjusting the extrusion multiplier in your slicer settings to fine-tune the amount of filament being extruded.
  4. Use a different slicer: Itā€™s possible that the issue lies with your slicer settings. Try using a different slicer, such as PrusaSlicer or Simplify3D, to see if that helps.
  5. Check your filament: Ensure that your filament is high quality and stored correctly, away from moisture and dust. Low-quality filament or improperly stored filament can lead to inconsistent results.
  6. Experiment with print temperature and speed: Try printing at different temperatures and speeds to see if that makes a difference. Some filaments, such as PETG, may require higher temperatures to print consistently.
  7. Consider upgrading your hotend: If all else fails, you may want to consider upgrading your hotend to a higher quality one, such as an E3D v6 or a Micro Swiss all-metal hotend.